SAXS Sample Prep


  • Macromolecule sample and exact buffer
  • The sample cell takes 15 ul. 20 ul is safer when using the robot.
  • Minimum concentration is 1 mg/ml
  • Maximal concentration is 10 mg/ml
  • Identical buffer required (>5ml recommended)

Please download and fill out this mandatory shipping form before sending your samples. It should be included with all samples sent to the SIBYLS beamline for data collection.

Send samples to:

Lawrence Berkeley Lab
1 Cyclotron Road
MS 6R2100
Berkeley, CA 94720
ATTN:  Kathryn Burnett / 12.3.1

Send an email to Kathryn Burnett with federal Express shipping information so she knows the sample is coming and so she can track the sample online.

RECOMMENDATIONS for Sample preparation ( for routine analysis and first-time users )


We recommend not using detergents in entire protein preparation. We observed only a few cases (~1%) where the detergent ( below CMC concentration ) did not affect the protein signal.


The most common problem at the beamline is aggregation in the sample. Since larger particles scatter X-rays more strongly than small particles (albeit less than visible light scattering), aggregation will bias the results. We strongly recommend doing either DLS, native gel, or gel filtration (best). If your sample has a tendency to aggregate over time, it is possible to prepare the sample at dilute concentrate and then concentrate just prior to data collection.


The higher the concentration, the better the signal. However, there's a balance between problems with aggregation/oligomerization at the higher concentrations, unless the macromolecule is well-behaved. We generally recommend 1-5 mg/ml and doing a concentration series. If we are given 50 ul, we can do serial dilutions. Homo-oligomers can give okay signal down to 1 mg/ml. Ideally data should be collected on at least three different concentration of the macomolecule in the range 1-10 mg/ml to identify any concentration dependent behavior. Aggregation precludes data analyses. Concentration determination: OD280 is considered the most accurate although buffer subtraction is very important (oxidized DTT has an absorbance at 280). We have a nanodrop at the beamline (2 ul sample vol), and we recommend taking conc. right after removal from dialysis. Please verify that your conc. is correct--overestimation is a common problem.


96 well plate sample format. Our beamline operates with a pipetting robot, which works with 96 well plates. The samples can be shipped in fast frozen state in the single concentration with minimal volume 24uL . 


We strongly recommend dialyzing sample. The difference between the scatter of the macromolecule and buffer is so low, that simply making up the "equivalent" buffer is not sufficient to get accurate subtraction. To retain sample volume, we have found that the Hampton dialysis buttons (eg. 30-50 ul size) are ideal at keeping the volume constant. If you haven't used the buttons before, practice with saran wrap. Hampton has a nice set of instructions--many beginners fine the golf tee to make the difference. To remove the sample, we use a Hamilton syringe with a blunt needle, pierce the dialysis membrane, and remove the sample. We usually lose only a couple ul. The dialysis buttons fit well into 50 ml tubes--I have crammed in 7-8 in the same tube. The 50 ul volume allows us to do serial 1:1 dilutions. The buffer in the concentrator flow through is also good for subtraction (depending on your buffer components).


Salt increases the background, but we've gotten good signal with up to 1 M salt. Concentration of the macromolecule has more of an impact on signal than the buffer, so if the sample is monodisperse in high salt, put it in high salt. 1-5% glycerol cuts down quite a bit on radiation damage, although we've also seen oligomerization induced by addition of glycerol. If you have the quantity available, trying several different buffers is recommended. For a start, ideal buffer might contain 100-200 mM salt with 5% glycerol. Please send at least 15 mL of buffer with your sample.


This happens, maybe 1 in 10 samples we observe changes in the SAXS scattering curve due to radiation damage. We've found that complexes do seem more stable, so if you have a complex, then throw that in as well. Glycerol is a pretty good radical scavenger, so 5% glycerol is good.


The beamline is equipped with Peltier that can vary the temperature from 0 deg to 50 deg.


If you are sending in your sample for data collection as a collaborative effort with a beamline scientist, please

    • 1. Label tubes uniquely so that when they are stuck in the frig with many other tubes, they don't get lost. Recommend your initials with a number (eg st1) and date of sample.
a. Sample
b. Corresponding buffer
    • 2. Send information on the sample that will help during data collection:
a. MW and oligomerization state, if known
b. Number of residues
c. Extinction coefficient 280
d. Buffer components
e. Sample concentration done on concentrated sample and what method used.


Sample cell has flat Mica windows. We generally keep our Pilatus 2M detector at 1.5 m distance.

We have developed a time slicing mode for collection. Each sample is collected multiple times with the same exposure length. We generally collect every .3 seconds for a total of 10 seconds resulting in 30 frames per sample (though there is some variation depending on beam optimization).

The time of the short exposure is selected to minimize detector overloads near beam stop but to provide accurate measurement close to the beam stop. The long exposure is generally ten-fold longer to accurately measure high angle data.

We can change the X-ray wavelength to optimize for sample size, although 11 to 13 KeV is generally fine for most samples under 40-150 kD MW. Qmin should be collected at an angle < pi/dmax. Detector parameters however must be calculated for each wavelength (we use silver behamate), so if you want to change the wavelength, please advise the beamline scientist.


1. Circular integration, normalization, and subtraction of sample and buffer image files. Usage --> ogreNew sampleimg bufferimg

ogreNew *005 *004

[subtracts image4 (buffer) from image5 (sample) and generates I vs S scattering curve]

2. We use xmgrace to visualize the scattering curves.

ivq sampleimg.dat

3. Preliminary analysis using PRIMUS and Gnom during data collection (from Svergun's group) is extremely helpful in determining the quality of your data. If your data is not good, subsequent analysis will be inaccurate.

PRIMUS-guinier analysis indicates aggregated sample (increasing and non-linear slope towards I0). I0 estimate also good for estimating molecular weight using known standards (MW is linear with I0/conc).

PRIMUS-porod analysis gives rough estimate of molecular weight (porod volume*1.2/2 ~ molecular weight).

PRIMUS--SASPLOT (I*S^2 vs S) gives indication of folded nature of molecule. For example, well folded proteins show parabolic curve that returns to baseline. Unfolded or random coil molecules will increase with angle. Please note that too dilute sample and poor signal at high S or inaccurate background subtraction will also give this result. GNOM--p(r) analysis. Aggregated sample will have unstable Rg and p(r) curve will have small bump close to Dmax.